Josh Warner

@michaelaye, there has been some brief chatter about NaNs but no real impetus to make it happen. I think a large part of this is twofold: - Unlike other fields...

@michaelaye Indeed, those would be an ideal place to begin! Simply omitting NaNs seems perfectly appropriate for non-adaptive thresholding (adaptive may be a little trickier).

@paalge This looks very interesting, thanks for the PR! After having a brief look, here's a short to-do list that needs to happen before getting merged in: - As this...

@paalge possibly true for research, but scikit-image is BSD licensed and may be used for profit. When you start distributing code there is a higher standard. Usually "clean room" implementations...

I agree with @jni, pertaining to the distinction between a rank-3 array intended to represent a 3D volume and a rank-3 array which represents channels or a collection of images...

@kne42 you're right. That's what I was referencing. However, it turns out there are real performance gains when iterating over the _last_ axis due to the way arrays are allocated...

Should be forward compatible now, please check against scikit-fuzzy 0.5.0 just released @jcreinhold.

@dwillis Most 2020 elections are now using the `web.XXXXXX` URL format, for example the GA 2020 general election here: https://results.enr.clarityelections.com/GA/105369/web.264614/#/summary Installing from source current as of posting, neither this link...

Completely agree. This has always bothered me, too. The dictionary is almost certainly the right API choice. An OrderedDict might be the best choice, actually, to maintain the iteration order...

@ahojnnes :+1: agreed, if possible the ideal change would be a drop-in replacement. That function gets used directly or indirectly quite a bit, so existing test coverage should be very...